A sample truncation rule for product reliability estimation with multiple censoring data

For product reliability estimation in the condition of only some successfully operation times available, this paper presents an approach to estimate the location parameter of three-parameter Weibull distribution, as well as a criterion to construct an effective data set, i.e. to truncate the original sample into a beneficial sample of smaller size. Through comparing the estimated life distributions and reliabilities based on different sub-sets of the original one composed of all the available operation time records, a criterion is worked out to truncate the original sample of the censoring life data for reliability estimation. Lots of simulated data are generated by Monte Carlo sampling, by which the proposed criterion is testified.